Content uploaded by Benjamin Slater
Author content
All content in this area was uploaded by Benjamin Slater on Sep 04, 2023
Content may be subject to copyright.
A translative study of context-dependent
sequence learning in rats and common marmosets
Benjamin Slater 1,2, Emma Woolgar 3, Jennifer Nacef 1, Christopher Petkov 1,4, Yukiko Kikuchi 1 & Alexander Easton 2
1 Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK
2 Department of Psychology, Durham University, Durham, UK
3 Comparative Biology Centre, Newcastle University, Newcastle upon Tyne, UK
4 Department of Neurosurgery, University of Iowa, Iowa, USA
Rodent task design
******
Introduction
Summary of findings
Two-context sequencesSingle-context sequences
8 Lister Hooded rats (Rattus norvegicus)v
Threshold performance = 10/12 trials correct across two consecutive days
Sequence learning has been studied in both rodents and primates, however,
complexity of the sequences learned differs greatly between the two. For rodents,
focus has been on object and item associations in a more immersive environment1,
whereas in primates, more complicated abstract decision-making tasks are utilised
requiring selection of multiple objects2, usually on a touchscreen/virtual environment3.
We aim to bridge this gap by creating a task that can be run analogously across
rodents and primates using objects in a 3D maze to create overlapping sets of
sequences. Complexity of the sequences was similar for both species, with object
identity and location remaining the same across multiple environments requiring
each animal to use implicit cues to determine the correct sequence order.
3 Common marmosets (Callithrix jacchus)
Threshold performance = 30/40 trials correct across two testing days
6/8 rats learned single-context
sequences but only 2/6 rats learned
sequences across two contexts
All 3 marmosets learned single-
context sequences but only 2 learned
sequences across two contexts
Alternating two-context sequences in
a session proved difficult to learn
despite prior successes
Rats that learned all sequence types
were the quickest to learn throughout
the entire experiment
Greater error ratio for two-context
sequences when context was the
same as the trial before
Greater error ratio for single-context
sequences when context differed to
the trial before
Objects present
along x-axis
Objects present
along y-axis
Both rats and marmosets are capable
in learning overlapping sequences that
involve making decisions dependent
on their surrounding context.
Rats made the majority of their errors
when choosing the second element of
the sequence, especially during two
context sequences that involved a
transition between different contexts.
Comparing trial-by-trial performance
showed contextual information from
the previous trial affected each rat's
decision during their next trial.
Marmosets were quick to learn both
sequences but lacked motivation to
reach threshold for more complicated
two-context sequences, favouring to
leave the apparatus mid-trial instead.
***
n.s
2-CTX
Marmoset task results
***
Error rates for rats that learned
two-context temporal sequences did
not differ between error types
Failure to learn two-context
sequences involved more errors
selecting the second object
n.s
Rodent task results
Threshold = 10/12 trials correct
Threshold = 10/12 trials correct
***
x
Comparing context change across trials
Comparing successful vs unsuccessful rats
Comparing errors for individual marmosets
Marmoset task design
Single-context sequences Two-context sequences
b.j.a.slater2@newcastle.ac.uk www.benslaterneuro.com
References: 1. Navawongse & Eichenbaum, J. Neuro. (2013) 2. Mansouri, Freedman & Buckley, Nat. Rev. Neuro. (2020) 3. Gaffan, D., J. Cog. Neuro. (1994)
First Object Error Sequence Error First Context Incomplete Second Context Incomplete
n.s ***
Each marmoset showed understanding of two-context sequences but success
was negatively affected by lack of motivation to alternate between contexts
ALT-CTX
30/40 trials correct
30/40 trials correct
10/12 trials correct